Perspectivas. Se debe tener cuenta que el Cuadro de Mando Integral debe servir como herramienta para ser utilizado como un sistema de

sticking to the rules. The required level of discipline is achieved when the proper habits are formed.

At first glance, 5S may seem to be nothing but an office-cleaning exercise, but in reality, these are principles to be followed in designing new processes and workplace layouts. The 5S principles can help reduce non-value-adding activity by as much as 25 percent. That is time that can be allocated to revenue-generating activities.

Line Balancing

Line balancing and takt time are related concepts. Line balancing involves distributing the work content of the various operations in a process in such a way that the time required for each operation is equal to or less than the takt time of the line. Line balancing is used to:

1. Avoid a buildup of work in progress.

2. Begin identifying the number of resources required to run the “line.”

This ensures the optimal usage of personnel, while meeting customer demand. Figure 4.4 outlines the benefits of line balancing.

Figure 4.4 Line Balancing

As outlined in Figure 4.4, there are 12 distinct steps in the process. The calculated takt time required to meet customer demand is 12 minutes. This means that every step has to complete its required task within 12 minutes. While some steps can complete their tasks in far less than 12 minutes (which can be translated as having excess capacity), others, like Step 5, exceed that requirement. Any process step that exceeds takt time will act as a bottleneck.

The graphic demonstrates that through consolidation of process steps and reallocation of work, resources can be better utilized, there are fewer steps, and each step is capable of meeting takt time.

There are four distinct strategies for overcoming line-balancing challenges:

1. Reduce non-value-adding activities in process steps where the cycle time exceeds takt time.

2. Relocate the work, grouping tasks into segments of work equal to or less than takt time.

3. Use overtime.

4. Add resources or additional equipment.

Figure 4.5 shows how the elimination of non-value-adding activities helps bring balance back to the process.

Figure 4.5 Process Improvement

Elimination or Reduction of Non-Value-Adding Activities

As discussed in the takt time and 5S sections, one of the main goals of a kaizen event is to improve capacity and process capability by eliminating non-value-adding activities. The upcoming example is from a brokerage firm, where a team is focused on improving the performance of the wire transfer process. A common process in most financial institutions, this is a very time-sensitive activity that involves great risk for both the institution and the client who is requesting the transfer (wrong amount, wrong recipient, or unavailable funds).

Once the team developed a process map of the “as-is” state, it leveraged the kaizen event to

gain consensus on which activities were non-value-adding activities, prioritize them, and reengineer the process. Figures 4.6 and 4.7 show the before and after picture.

Figure 4.6 Process Before Kaizen

Figure 4.7 Process After Kaizen

The results after a two-day event were:

• Reduced the number of steps from 14 to 5

• Eliminated four inspection points

• Reduced the process rejection and rework by 85 percent

• Reduced the overall process cycle time by over 70 percent

Kaizen Summary

You can use a kaizen event to identify and implement solutions involving cycle time and standardization efforts. Here are some principles for this process:

• Ensure that all bottlenecks, inefficiencies, and waste have been identified prior to engaging in kaizen.

• Create a cross-functional team—all members who are affected by the process should be involved.

• Remove bottlenecks using line balancing:

Balance tasks by takt time.

Reduce transit time.

Reduce manual work time.

Improve workflow by using 5S and reducing waste.

Six Sigma Tools

Not all project goals can be met by using Lean principles. What if you are considering

changing service providers or vendors? How will you know if the new vendor’s performance is statistically better? What if you have found multiple factors that affect your process—how will you know which ones have the most statistical significance? The answers to these

questions can be attained by using Six Sigma tools, namely hypothesis testing (discussed in Chapter 3, “Analyze”) and regression (see Figure 4.8).

Figure 4.8 Lean and Six Sigma Tools

Simpler Statistics—Going Back to What We Know

While there are many statistical tools for the analysis and graphical display of data, the

objective of Six Sigma is to use whatever is appropriate to get the required answers. And this means that sometimes there are simpler principles that can yield the required findings. The objective of all Green Belts and Black Belts is to keep it statistically simple (KISS). To

demonstrate this point, a regional bank was considering switching its checkbook issuance vendor. Based on historical performance, it knew that the time of year and the type of checkbook the client requested affected the delivery cycle time. It asked the new vendor under consideration (Vendor B) to provide information on its delivery cycle time for a period of two years. Using hypothesis testing, the bank could then determine whether its existing vendor (Vendor A) was performing better than or the same as the new vendor under

consideration. Here is how the bank set up the test: for the null hypothesis, it assumed that the average delivery cycle time was the same for both vendors, and for the alternative

hypothesis, it assumed that the cycle times of the two vendors were different.

Figure 4.9 shows the Minitab box plot and statistical analysis output from hypothesis testing.

Figure 4.9 Box Plot for Hypothesis Testing

Each vendor has provided more than 20,000 data points (denoted by the column labeled N), and the average delivery time for Vendor A is 38 days compared with 33 days for Vendor B. Since we have a p value less than 0.05, we can conclude that we need to reject the null hypothesis. Therefore, there is a statistical difference between Vendor A and Vendor B. Based on the box plot, Vendor B does perform better.

The next step for the bank is to consider the following items before making the switch and piloting the new vendor:

• Is the vendor qualified?

• Are there practical issues (location, other CTQs)?

• Is one vendor easier to work with?

• What will be the impact on other processes?

• What will be the impact on cost?

The use of hypothesis testing, a common yet powerful Six Sigma tool, provided the team with critical information on whether to proceed with the project or abandon the idea of making the switch.

Introduction to the Idea of Piloting Solutions

A pilot involves testing a process improvement on a small scale in a real business

environment, whenever feasible. The objective of the pilot is to collect and analyze real process data in order to:

• Confirm that your proposed solution will achieve the targeted performance (such as increasing production or reducing defects).

• Identify any potential implementation problems (technology, training, and so on) prior to full-scale implementation.

If piloting the improvement is an option, then the team needs to define what a realistic environment looks like. The team needs to ensure that the conditions of the pilot reflect the true business conditions. That is, the new improvement should really experience all the possible process variations (region, shift, request type, or product type). A sample list of questions to consider when setting up a pilot is:

1. Should the pilot be conducted in one region or many? If many, should the pilots in the different regions be done simultaneously or one by one?

2. How long should the pilot go on?

3. Should weekends, holidays, and/or peak periods be included?

4. Who should participate? Should it be the best, worst, or average workers?

5. Which product types should be included?

6. How will the process data be collected?

If a pilot is done correctly, the benefits that will be attained include:

• Better understanding of the effects of your solution on the organization and the customer

• Proper planning for a successful full-scale implementation

• The ability to release an early version of your solution to a particular market segment that has an urgent need for the change

• Lowering your risk of failing to meet your improvement goals when the solution is fully implemented

• More accurate prediction of the monetary savings resulting from your solution

• Justification of the investment required for full-scale implementation

• Identification of potential problems with the solution implementation on a larger scale

Linear Regression

Do missing data points on an application affect loan processing cycle time? Is there a

relationship between the number of patients that need to be admitted to the ward and wait time? We can use graphical tools like a scatter plot to demonstrate a relationship pictorially,

but is the relationship statistically significant? Figure 4.10 is a scatter plot of the number of missing data points on a new loan application and the cycle time required to complete the

“documentation” step (in which all the required customer data are compiled and the application is considered ready for underwriting) in loan processing.

Figure 4.10 Scatter Plot of Missing Data vs. Cycle Time

There seems to be a positive relationship between missing customer data and

documentation cycle time; that is, as the number of missing data points on an application increases, so does the processing cycle time in the back office.

Regression analysis helps you further explore this relationship by building a mathematical equation that links the process Xs (for example, missing customer data) to the process

output Y (for example, loan application processing cycle time). Regression is a statistical technique that is used to model and investigate the relationship between two or more

variables. The model is often used for prediction: if there are three missing data points on an application, what is the expected cycle time to complete the documentation? This type of analysis begins with the team first selecting the process output, or Y, that it wants to predict

—is it process cycle time, lost deals, or calls abandoned? Then the team needs to agree on a list of independent variables, or Xs, that it wants to use as predictors—for example, number

of agents, applications, and missing data points. The fishbone tool and the FMEA are perfect tools for generating a list of potential Xs.

Using the same data, a regression plot can be generated using statistical software. The regression plot in Figure 4.11 has two key outputs:

Figure 4.11 Regression Plot

• A mathematical equation linking documentation cycle time (our process output, Y) to missing data points (process input, X)

• R2 value

The closer the R2 value is to 100 percent, the better the prediction capability of the

equation. In statistical terms, it is the fraction of the variation in the output variable that is explained by the equation. So in this case, 89 percent of the variation that we see in the documentation cycle time can be attributed to missing data in the application; the other 11 percent is affected by other variables—errors in data collection, volume of applications on the days when the data were collected, the field rep filling out the application, and so on.

So how do we use this equation? Assume that we have a documentation cycle time target

of 40 minutes. We can now determine the maximum number of missing data points that an application can have and still meet the cycle time requirements.

Therefore, any application that is submitted with more than four missing data points has a high probability of missing the 40-minute processing cycle time.

What if there appears to be no relationship between your Y and X—that is, the R2 value is very low?

That would mean that, practically speaking, the X does not influence or change your Y.

More process data may not necessarily result in a relationship. Some things to consider before you abandon your data are:

• Review your data with a process expert to make sure that they make sense.

• Review your process map in conjunction with your data collection plan. Are your data hiding multiple processes, such as collecting data from multiple teams, products, or regions? When the data were being collected, combining these effects may not have seemed critical, but perhaps the lack of correlation between X and Y indicates that these effects do heavily influence the process output. As an example, if the teams work

differently, or if products are sold via different channels, the outputs may be different (for example, process cycle time between teams or channels will be different), and that can affect the results of your regression. In cases where the data may reflect multiple processes, each of these processes needs to be examined individually.